This paper investigates 'Reactive Tokens' in Mandarin Chinese, Japanes
e, and English. Our definition of 'Reactive Token' (= 'RT') is 'a shor
t utterance produced by an interlocutor who is playing a listener's ro
le during the other interlocutor's speakership'. That is, Reactive Tok
ens will normally not disrupt the primary speaker's speakership, and d
o not in themselves claim the floor. Using corpora of conversational i
nteractions from each of the three languages of our study, we distingu
ish among several types of RTs, and show that the three languages diff
er in terms of the types of RTs favored, the frequency with which RTs
are used in conversation, and the way in which speakers distribute the
ir RTs across conversational units.